feat/8387-search-with-acronyms

server/src/main/java/au/org/aodn/ogcapi/server/core/model/enumeration/StacBasicField.java

@@ -21,7 +21,6 @@ public enum StacBasicField {
             "summaries.organisation_vocabs"
     ),
     Links("links", "links"),
-    LinksTitle("links.title", "links.title"),
     LinksAiRole("links.ai:role", "links.ai:role"),
     Collection("collection", "collection", "collection.keyword"),
     AssetsSummary("assets", "assets"),

server/src/main/java/au/org/aodn/ogcapi/server/core/service/ElasticSearch.java

@@ -286,20 +286,16 @@ public ElasticSearchBase.SearchResult<StacCollectionModel> searchByParameters(Li
                     else {
                         should.add(CQLFields.fuzzy_title.getPropertyEqualToQuery(term));
                         should.add(CQLFields.fuzzy_desc.getPropertyEqualToQuery(term));
+                        // Phrase match for acronym-synonym support (ticket #8387): when an acronym
+                        // is expanded into its multi-word full name, match_phrase requires those
+                        // words to appear consecutively, alongside (not replacing) fuzzy matching.
+                        should.add(CQLFields.phrase_title.getPropertyEqualToQuery(term));
+                        should.add(CQLFields.phrase_desc.getPropertyEqualToQuery(term));
                     }
                     should.add(CQLFields.parameter_vocabs.getPropertyEqualToQuery(term));
                     should.add(CQLFields.organisation_vocabs.getPropertyEqualToQuery(term));
                     should.add(CQLFields.platform_vocabs.getPropertyEqualToQuery(term));
                     should.add(CQLFields.id.getPropertyEqualToQuery(term));
-                    // A request to not using acronym in title and description in metadata, hence these
-                    // acronym moved to links, for example NRMN record is mentioned in the link title.
-                    // This is a work-around to the requirement but still allow use of NRMN
-                    // links_title_contains and credit_contains use match query by default, exact match is not applied here
-                    // links_title_contains weighted lower as it may contain combined title+description content
-                    should.add(BoolQuery.of(b -> b
-                            .should(CQLFields.links_title_contains.getPropertyEqualToQuery(term))
-                            .boost(0.5f)  // lower boost to reduce promotion of link-title-only matches
-                    )._toQuery());
                     should.add(CQLFields.credit_contains.getPropertyEqualToQuery(term));
                 }
             }

server/src/test/java/au/org/aodn/ogcapi/server/common/RestApiTest.java

@@ -128,27 +128,30 @@ public void verifyApiCollectionsQueryOnText2() throws IOException {
                 collections.getBody().getCollections().get(1).getId(),
                 "Correct UUID - 9fdb1eee-bc28-43a9-88c5-972324784837");
     }
-    /**
-     * Acronym is not encourage to use in title or description, so NRMN record is not found, the acronym usually
-     * appears in links title, this test is make sure NRMN record is found from link as well.
-     * @throws IOException - IO Exception
-     */
+
+    /** Searching an acronym ("NRMN") matches records that only contain its full form ("National Reef Monitoring Network"). */
     @Test
-    public void verifyApiCollectionsQueryOnText3() throws IOException {
+    public void verifyAcronymSynonymSearch() throws IOException {
         super.insertJsonToElasticRecordIndex(
-                // This is NRMN record where word NRMN not in title/desc but links
-                "8cdcdcad-399b-4bed-8cb2-29c486b6b124.json",
-                "7709f541-fc0c-4318-b5b9-9053aa474e0e.json"
+                "acronym_demo_only_acronym.json",
+                "acronym_demo_only_fullname.json",
+                "acronym_demo_unrelated.json"
         );
 
-        // Call rest api directly and get query result
-        ResponseEntity<ExtendedCollections> collections = testRestTemplate.getForEntity(getBasePath() + "/collections?q=NRMN", ExtendedCollections.class);
-        assertEquals(1, Objects.requireNonNull(collections.getBody()).getTotal(), "Only 1 hit");
+        // Search the acronym -> should hit B (the full-name-only record) via synonym expansion.
+        ResponseEntity<ExtendedCollections> byAcronym = testRestTemplate.getForEntity(
+                getBasePath() + "/collections?q=NRMN", ExtendedCollections.class);
+        assertEquals(1,
+                Objects.requireNonNull(byAcronym.getBody()).getTotal(),
+                "Searching 'NRMN' should find the full-name-only record via synonym expansion"
+        );
         assertEquals(
-                "8cdcdcad-399b-4bed-8cb2-29c486b6b124",
-                collections.getBody().getCollections().get(0).getId(),
-                "Correct UUID - 8cdcdcad-399b-4bed-8cb2-29c486b6b124");
+                "acdemo02-0000-0000-0000-000000000002",
+                byAcronym.getBody().getCollections().get(0).getId(),
+                "The matched record should be the full-name-only fixture (B)"
+        );
     }
+
     /**
      * The datetime field after xxx/.. xxx/ etc. It uses CQL internally so no need to test Before After During in CQL
      */
@@ -565,7 +568,7 @@ public void verifyCQLPropertyScore() throws IOException {
 
         // Lower score but the fuzzy is now with operator AND, therefore it will try to match all words 'dataset' and 'includes' with fuzzy
         collections = testRestTemplate.getForEntity(getBasePath() + "/collections?q='dataset includes'&filter=score>=1", Collections.class);
-        assertEquals(3, Objects.requireNonNull(collections.getBody()).getCollections().size(), "hit 1, with score 3");
+        assertEquals(1, Objects.requireNonNull(collections.getBody()).getCollections().size(), "hit 1, with score 3");
         assertEquals("bf287dfe-9ce4-4969-9c59-51c39ea4d011", Objects.requireNonNull(collections.getBody()).getCollections().get(0).getId(), "bf287dfe-9ce4-4969-9c59-51c39ea4d011");
 
         // Increase score will drop two record

server/src/test/java/au/org/aodn/ogcapi/server/features/RestApiTest.java

@@ -244,7 +244,7 @@ public void verifyCorrectPageSizeDataReturnWithQuery() throws IOException {
                 "Record return size correct"
         );
         // Total number of record should be this
-        assertEquals(5, collections.getBody().getTotal(), "Get total works");
+        assertEquals(4, collections.getBody().getTotal(), "Get total works");
 
         // The search after give you the value to go to next batch
         assertEquals(3, collections.getBody().getSearchAfter().size(), "search_after have three values");
@@ -273,7 +273,7 @@ public void verifyCorrectPageSizeDataReturnWithQuery() throws IOException {
                 "Record return size correct"
         );
         // Total number of record should be this as the same search criteria applies
-        assertEquals(5, collections.getBody().getTotal(), "Get total works");
+        assertEquals(4, collections.getBody().getTotal(), "Get total works");
 
         // The search after give you the value to go to next batch
         assertEquals(3, collections.getBody().getSearchAfter().size(), "search_after have three values");
@@ -302,7 +302,7 @@ public void verifyCorrectPageSizeDataReturnWithQuery() throws IOException {
                 "Record return size correct, returns the 3 remaining matching docs"
         );
         // Total number of record should be this as the same search criteria applies
-        assertEquals(5, collections.getBody().getTotal(), "Get total works");
+        assertEquals(4, collections.getBody().getTotal(), "Get total works");
 
         // The search after give you the value to go to next batch
         assertEquals(3, collections.getBody().getSearchAfter().size(), "search_after three fields");
@@ -370,7 +370,7 @@ public void verifyCorrectPageSizeAndScoreWithQuery() throws IOException {
                 "Record return size correct"
         );
         // Total number of record should be this
-        assertEquals(5, collections.getBody().getTotal(), "Get total works");
+        assertEquals(4, collections.getBody().getTotal(), "Get total works");
 
         // The search after give you the value to go to next batch
         assertEquals(3, collections.getBody().getSearchAfter().size(), "search_after three fields");
@@ -414,7 +414,7 @@ public void verifyCorrectPageSizeAndScoreWithQuery() throws IOException {
                 "Record return size should be 3 or 4 (bc55eff4 borderline), got: " + returnedSize);
 
         // Total number of record should be this as the same search criteria applies
-        assertEquals(5, collections.getBody().getTotal(), "Get total works");
+        assertEquals(4, collections.getBody().getTotal(), "Get total works");
 
         // The search after give you the value to go to next batch
         assertEquals(3, collections.getBody().getSearchAfter().size(), "search_after three fields");

server/src/test/java/au/org/aodn/ogcapi/server/service/ElasticSearchTest.java

@@ -152,18 +152,16 @@ public void searchByParametersWithDoubleQuote() {
             } else {
                 should.add(CQLFields.fuzzy_title.getPropertyEqualToQuery(term));
                 should.add(CQLFields.fuzzy_desc.getPropertyEqualToQuery(term));
+                should.add(CQLFields.phrase_title.getPropertyEqualToQuery(term));
+                should.add(CQLFields.phrase_desc.getPropertyEqualToQuery(term));
             }
             should.add(CQLFields.parameter_vocabs.getPropertyEqualToQuery(term));
             should.add(CQLFields.organisation_vocabs.getPropertyEqualToQuery(term));
             should.add(CQLFields.platform_vocabs.getPropertyEqualToQuery(term));
             should.add(CQLFields.id.getPropertyEqualToQuery(term));
-            should.add(BoolQuery.of(b -> b
-                    .should(CQLFields.links_title_contains.getPropertyEqualToQuery(term))
-                    .boost(0.5f)  // lower boost to reduce promotion of link-title-only matches
-            )._toQuery());
             should.add(CQLFields.credit_contains.getPropertyEqualToQuery(term));
         }
-        assertEquals(8, should.size(), "Exact match should produce 8 queries (title + description + other fields)");
+        assertEquals(7, should.size(), "Exact match should produce 7 queries (title + description + 5 other fields)");
         assertTrue(should.get(0).isMatchPhrase(), "Title query should be MatchPhraseQuery");
         assertTrue(should.get(1).isMatchPhrase(), "Description query should be MatchPhraseQuery");
     }
@@ -182,18 +180,16 @@ public void searchByParametersWithoutDoubleQuote() {
             } else {
                 should.add(CQLFields.fuzzy_title.getPropertyEqualToQuery(term));
                 should.add(CQLFields.fuzzy_desc.getPropertyEqualToQuery(term));
+                should.add(CQLFields.phrase_title.getPropertyEqualToQuery(term));
+                should.add(CQLFields.phrase_desc.getPropertyEqualToQuery(term));
             }
             should.add(CQLFields.parameter_vocabs.getPropertyEqualToQuery(term));
             should.add(CQLFields.organisation_vocabs.getPropertyEqualToQuery(term));
             should.add(CQLFields.platform_vocabs.getPropertyEqualToQuery(term));
             should.add(CQLFields.id.getPropertyEqualToQuery(term));
-            should.add(BoolQuery.of(b -> b
-                    .should(CQLFields.links_title_contains.getPropertyEqualToQuery(term))
-                    .boost(0.5f)  // lower boost to reduce promotion of link-title-only matches
-            )._toQuery());
             should.add(CQLFields.credit_contains.getPropertyEqualToQuery(term));
         }
-        assertEquals(8, should.size(), "Fuzzy match should produce 8 queries");
+        assertEquals(9, should.size(), "Fuzzy match should produce 9 queries (fuzzy + phrase title/desc + 5 other fields)");
         assertTrue(should.get(0).isMatch(), "fuzzy_title should be MatchQuery");
     }
 }

server/src/test/resources/databag/acronym_demo_only_acronym.json

@@ -0,0 +1,17 @@
+{
+  "id": "acdemo01-0000-0000-0000-000000000001",
+  "title": "NRMN Reef Survey Data 2024",
+  "description": "Reef benthic cover survey data collected at standard NRMN sites along the Australian coast.",
+  "extent": {
+    "bbox": [[110.0, -45.0, 155.0, -10.0]],
+    "temporal": [["2024-01-01T00:00:00Z", "2024-12-31T23:59:59Z"]]
+  },
+  "summaries": {
+    "score": 50,
+    "status": "completed",
+    "scope": { "code": "dataset", "name": "Demo - only acronym in title/description" },
+    "parameter_vocabs": [],
+    "platform_vocabs": [],
+    "organisation_vocabs": []
+  }
+}

server/src/test/resources/databag/acronym_demo_only_fullname.json

@@ -0,0 +1,17 @@
+{
+  "id": "acdemo02-0000-0000-0000-000000000002",
+  "title": "National Reef Monitoring Network Sub-Facility Survey",
+  "description": "Long-term observations of reef benthic and fish communities conducted by the National Reef Monitoring Network across temperate and tropical Australia.",
+  "extent": {
+    "bbox": [[110.0, -45.0, 155.0, -10.0]],
+    "temporal": [["2024-01-01T00:00:00Z", "2024-12-31T23:59:59Z"]]
+  },
+  "summaries": {
+    "score": 50,
+    "status": "completed",
+    "scope": { "code": "dataset", "name": "Demo - only full name in title/description" },
+    "parameter_vocabs": [],
+    "platform_vocabs": [],
+    "organisation_vocabs": []
+  }
+}

server/src/test/resources/databag/acronym_demo_unrelated.json

@@ -0,0 +1,17 @@
+{
+  "id": "acdemo03-0000-0000-0000-000000000003",
+  "title": "Ocean Temperature Observations off Tasmania",
+  "description": "Hourly sea surface temperature measurements collected by moored buoys.",
+  "extent": {
+    "bbox": [[143.0, -45.0, 150.0, -40.0]],
+    "temporal": [["2024-01-01T00:00:00Z", "2024-12-31T23:59:59Z"]]
+  },
+  "summaries": {
+    "score": 50,
+    "status": "completed",
+    "scope": { "code": "dataset", "name": "Demo - unrelated negative sample" },
+    "parameter_vocabs": [],
+    "platform_vocabs": [],
+    "organisation_vocabs": []
+  }
+}